Available online at www.sciencedirect.

com

ScienceDirect
Procedia Computer Science 95 (2016) 297 – 302

Complex Adaptive Systems, Publication 6

Cihan H. Dagli, Editor in Chief
Conference Organized by Missouri University of Science and Technology
2016 - Los Angeles, CA

Time-Frequency Spatial Wavelet Phase Coherence Analysis of EEG

in EC and EO During Resting State
Lal Hussain a* , Wajid Aziz b
a,b
University of Azad Jammu and Kashmir, Muzaffarabad, Pakistan
b
University of Jeddah, Kingdom of Saudi Arabia

Abstract

The electrophysiological brain activities are nonlinear in nature as measured by Electroencephalography (EEG). There are
coherent activities in brain not only seen during explicit tasks but also during rest. This article aims to investigate the Wavelet
based Phase coherence of oscillations of eye closed and eye open signals during resting states. The wavelet coherence is
computed for selected 19 electrodes according to 10-20 system from 129 channel EEG signals. The significance was obtained
using Wilcoxon Signed Rank test and pairwise wavelet coherence was computed for each possible combination. The Wavelet
Phase Coherence using Wavelet Transform gives significantly high results (P<0.05) in EC and EO signals during resting states in
frequency interval 0.5-50 Hz overall as well as in the band intervals such as delta (05-4 Hz), theta (4-7 Hz), alpha (7-13 Hz), beta
(13-22 Hz) and gamma (22-50 Hz).
©
© 2016 The
TheAuthors.
Authors.Published
Publishedbyby Elsevier
Elsevier B.V.
B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of scientific committee of Missouri University of Science and Technology.
Peer-review under responsibility of scientific committee of Missouri University of Science and Technology
Keywords: Wavelet Coherence; Resting State; Electroencephalography

1. Introduction

The nonlinear dynamics in normal resting state EEG are primarily concerned to study the dynamics in normal
EEG particularly in alpha rhythm. The previous studies [1] reveal that normal EEG are reflecting weak but

* Corresponding author. Tel.: +92-3345796278

E-mail address: lall_hussain2008@live.com

1877-0509 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of scientific committee of Missouri University of Science and Technology
doi:10.1016/j.procs.2016.09.338
298 Lal Hussain and Wajid Aziz / Procedia Computer Science 95 (2016) 297 – 302

significantly nonlinear structure (Gautama et al., 2003; Gebber et al., 1999; Maurice et al.,2002; Meyer-Lindenberg,
1996; Palus, 1996c; Pritchardet al., 1995a; Rombouts et al., 1995; Stam et al., 1999; Stepien, 2002). Previous studies
also reveal that alpha activity in EEG is dominant in normal individuals during eye-closed resting condition and
suppresses as visual stimulation. Alpha activity decreased during occipital regions (Berger, 1933; Adrian and
Matthews, 1934; Jasper, 1936; Smith, 1938) and also in posterior regions when the individuals open their eyes
(Chapman et al., 1962; Volavka et al., 1967; Legewie et al., 1969; Glass and Kwiatkowski, 1970; Gale et al., 1971).
These studies suggest that alpha desynchronization reflects the increased visual system functioning due to visual
stimulation being mediated by the Reticular Activating System (Volavka et al., 1967; Gale et al., 1971; Hardle et al.,
1984). (Deco et al., 2008, 2009; Ghosh et al., 2008 also found a surprising degree of complexity and coherent
activities in human brain during resting states EEG.
In this paper we aimed to investigate the nonlinear dynamics in EEG signals in healthy subjects with eye open
and close using Wavelet Phase coherence during resting state. Coherence measure is the phase difference between
two signals weighted by amplitude and measured over time in particular frequency. The values of coherence range
from 0 to 1, if the value of coherence is 1, it shows the complete agreement in phase difference, and 0 shows the
completely random phase difference. This study is aimed to study the EEG coherence by computing topographic and
frequency structure. For topographic information, the coherence will be computed over different location such as
inter-hemispheric and intra-hemispheric locations, local and global, sagittal and literal etc. in different frequency
ranges and relationships in different coherence pairs.

2. Material and Methods

2.1. Data Sets

EEG data with eye–closed (EC) and eye open (EO) during resting states comprising of 129 channels was
taken from Neurophysiological Biomarker Toolbox accessible from URL:
(https://www.nbtwiki.net/doku.php?id=tutorial:start#.VVXEKkaLXEs) as used and discussed in [2]. The Data
contains 16 healthy volunteers (ten females, median age: 27 years, range 21–48) with no history of neurological or
psychiatric illness, illegal substance abuse or use of psychotropic medication”.

2.2. Methods

In time-frequency plane, we can know the frequency of an oscillation at every point in time that is influenced by
time dependent component. Thus in phase space, the complexity and frequency domain may fail to interpret the
useful information of non-autonomous system [3]. The wavelet transform is of time series is then obtained by
convolving the complex wavelet of equation 1) with the time series at each scale s. i.e.
ஶ
ܹሺ‫ݏ‬ǡ ‫ݐ‬ሻ ൌ න Ȳ൫•ǡ ሺ–ǡ ɒሻ൯݂ሺ߬ሻ݀߬ሺͳሻ
ିஶ
The Morlet wavelet transform as defined in 1) was applied to EC and EO in resting states for 19 electrodes
according to 10-20 international system. We have used the central frequency of ߱௢ ൌ ʹ and rescaling in
increments of 5 % between s=0.5 and 80 Hz.
When there are two noisy signals interacting to each other with different spectra in power at same
frequency, their oscillation whether related or unrelated can be detected significantly over time using the phase
coherence.
Thus time-averaged wavelet phase coherence as defined in [4].

௅ ଶ ௅ ଶ
ͳ
ߨ௣  ൌ ඩቌ൝෍ •‹ሾ ߮ሺ‫ݐ‬௡ ሻሿൡ ൅ ൝෍ …‘•ሾ ߮ሺ‫ݐ‬௡ ሻሿൡ ቍሺʹሻ
‫ܮ‬
௡ୀଵ ௡ୀଵ

The wavelet values of equations (4) can be replaced with a normalized phasor as defined by [4].
The averaged phase coherence is defined as:
 Lal Hussain and Wajid Aziz / Procedia Computer Science 95 (2016) 297 – 302 299

௅ ௅ ଵȀଶ
ͳ ͳ
ߨ௣ ൌ ൥൭ ෍ ݁ ௜ఝభ ሺ௧೙ሻ ݁ ି௜ఝమ ሺ௧೙ ሻ ൱ ൈ ൭ ෍ ݁ ି௜ఝభ ሺ௧೙ ሻ ݁ ௜ఝమ ሺ௧೙ ሻ ൱൩ ሺ͵ሻ
‫ܮ‬௞ ‫ܮ‬௞
௡ୀଵ ௡ୀଵ

We can find ‫ܧ‬ሺߨ௣ ሻ to calculate autocorrelation and ‫ܩ‬௞ statistics using ܹ௞ ሺ‫ݐ‬௡ ሻ of the unit phasor for the new time
series.

3. Discussions

The coherence was computed for 19 electrodes as mentioned above for all possible combinations and resultant
matrices contains enormous information which required to be shown in form of topographic maps in one plot. We
have computed coherence for each frequency band such as delta, theta, alpha, beta and gamma for both eye-closed
and open during resting states. Each large circle in these figures shows the whole scalp where a small topographic
map is placed at each electrode position also repressing a scalp and contains the coherence with all possible
combination of electrodes.

(a) (b)

(c)

Fig. 1: (a) Wavelet coherence EC, (b) EO (c) Significance using paired test for each 19 electrodes in topographic maps. This figure contains 19
small topographic maps according to 10/20 international system inside a single head display. The map provide full picture of 129 electrodes of
EC subject during resting state in delta frequency band (red: EC>EO; blue: EC<EO; grey: No significance)
300 Lal Hussain and Wajid Aziz / Procedia Computer Science 95 (2016) 297 – 302

4. Discussions and Conclusion

The aim of present study is to investigate the temporal dynamics of resting state EEG using coherence and
topographic maps. We have computed wavelet coherence and examined in detail the topography and frequency
structure of EEG coherence during resting state in EC and EO subjects. The topographic changes have been
observed in delta, theta, alpha, beta and gamma bands however most significant differences were seen in alpha band
as reflected in Figure 1 and detailed results for other bands are shown in Table 1.

Table 1: Pairwise Wavelet Coherence with statistical significance (p<0.05) using Ranksum test
Prob Delta Theta Alpha Beta Gamma
C3 None O1 F8,Fp1*,P3,T7,P7 None O1
C4 None Cz* Cz*,F7,P4, T7 None None
Cz F7 C4*,F4* C4*,F3*,F4*,Fp1*,Fp2*,Fz*,O1* F4* T7,T8,P8
F3 T8* None Cz*,F4*,F8*,Fp2*,Fz*,O1*,P7* None None
F4 None Cz* Cz*,F3*,F7*,Fp1*,O2*,P4*,P8* Cz* None
F7 Cz None C4,F4*,F8*,Fp1*,Fp2*,Fz*,O1*,P3* None None
F8 None None C3,F3*,F7*,Fp1*,P4*,Pz*,P8* None None
Fp1 Fp2*,T8* Fp2*,T8* C3*,Cz*,F4*,F7*,F8*,Fp2*,T8*,P7* Fp2* Fp2*
Fp2 Fp1*,Fz*,P3*, Fp1* Cz*, F3*,F7*,Fp1*,Fz*,O2*,P7*,P8* Fp1*,Fz* Fp1*,Fz*,O1*
T7*,P7*
Fz Fp2* None Cz*,F7*,Fp1*,Fp2*,O1*,O2* Fp2* Fp2*
O1 None C3 Cz*,F3*,F7*,Fp1*,Fz* None C3,Fp2*
O2 None None F4*,F7*,F8*,Fp2*,Fz* None T8
P3 Fp2* None C3,F7* None None
P4 None None C4,F4*,F8* None None
Pz T8 None None None
F8*
T7 Fp2*,P8 None C3,C4,P7 None Cz,P8
T8 F3*,Fp1*,Pz Fp1* Fp1* Fp2 Cz,O2
P7 Fp2* None C3,F3*,Fp1*,Fp2*,T7 None None
P8 None None F4*,F8*,Fp2* None Cz, T7

Generally, coherence in EC subjects increases in some bands and decreases in other. Pairwise coherence
was computed excluding self-coherence in the form of topographic maps. The highest coherence was observed in
EC in frontal regions in alpha band. Whereas other bands such as delta, theta, beta also exhibit high coherence in
frontal regions in both EC and EO subjects and gamma shows high coherence in EC in Fz and Fp2 only. The
parietal regions also shows high coherence in Alpha and Beta bands for both EC and EO and few parietal regions
also shows coherent activity in delta and gamma bands. Most of the immediate adjacent electrodes show the higher
coherence at all frequency bands [5]. It was also observed overall that when the spatial distance between the
electrodes increases the coherence also decreases for both EC and EO at all frequency bands (In Table 1, the *
denote that where coherence of EC was higher than EO).
During EC, the highest pairwise coherence in alpha band was observed in frontal regions (F3-Fp1, F4-Fz,
Fp1-Fz, Fp2-Fz, Fz-Fp2), the other bands also shows high coherent activities in frontal (F3-fp1, F4-Fz,Fp1-F3, Fp2-
F4), occipital-parietal (O2-P8), parietal (P3-P7, P4-P8, P7-P3, P8-P4) and parieto-occipital (P7-O1) regions. A high
posterior (P3-P7, P7-P3) coherence is also shown in alpha band with EO.
In human, the oscillations with electric potential 8-12 Hz range with greater amplitude in posterior regions
usually recorded as sinusoidal waves and present 95% in healthy adults usually during eye-closed resting state [6].
In the neocortex this potential is believed to arise from the oscillation of postsynaptic potentials (Berger, 1929;
Cooper et al., 1965; Nunez et al., 2001). The alpha rhythm is considered as idle and diminish when the eyes are open
 Lal Hussain and Wajid Aziz / Procedia Computer Science 95 (2016) 297 – 302 301

or during mental activity. In alpha band, there is an apparent reduction in activity from eye closed to eye open
condition with no topographic change. The global alpha reduction activity is consistent with the previous EEG
studies reporting that across the scalp alpha desynchronization take place due to onset of visual stimulation Adrian
and Matthews, 1934; Jasper,1936; Smith, 1938; Chapman et al., 1962; Volavka et al.,1967; Legewie et al., 1969;
Glass and Kwiatkowski, 1970;Gale et al., 1971). Previous studies only revealed that entire cortex is activated in
response to visual stimulation due to desynchronization in alpha activities (Basar and Schurmannn, 1999) and have
not distinguished the arousal and activation activities in this way.
Delta activity [7] during early development dominates the human EEG and decreases during normal
development (John et al., 1980; Gasser et al., 1988; Harmony et al., 1990). Slow wave delta activity in development
is deemed important to support brain maturation and prune redundant cortical connections to exhibit positive
association between gray matter volume and delta activity (Whitford et al., 2007). Whereas the high wave delta
activity recorded during reading tasks is observed in dyslexic school age children (Spironelli et al., 2006; Penolazzi
et al., 2008; Spironelli and Angrilli, 2010) and dyslexic young adults (Rippon, 2000). However, increased delta and
theta activity is reported in children with reading and writing disabilities and in dyslexics in resting state EEG at the
age 9-18 years (Sklar et al., 1972; Colon et al., 1979; Pinkerton et al., 1989; Harmony et al., 1995).
The pathological and ordinary human brain functions investigated by [8] measuring the corticocortical
connectivity. (Thatcher et al., 1986; Tucker et al., 1986) estimated the coherence in frequency spectrum of EEG
studies in resting state subjects and measured the coupling functions across distinct brain regions of human brain.
(Scher et al., 1994) studied the coherence of various intrahemispheric and interhemispheric regions and found
greater EEG spectra patterns in preterm infants than in full-term infants which indicate the increased neuronal
connectivity due to extrauterine experience. (Lowe et al., 1998) investigated the coherence in left and right
hemisphere in resting states single and multi-slice echoplanner data in right and left hemisphere, precentral gyri
(motor cortex), calcarine fissures (visual cortex), and right and left hemisphere amygdalae. They observed low
frequency correlations in resting state data found in left/right symmetric functional cortices and assessed for whole
brain functional connectivity as well.
Several recent studies (Von 1999; Cisek and Turgeon 1999; Varela et al. 2001; Singer 1993; Womelsdorf and
Fries 2007) show that brain contains disparate and multiple neural activities in underlying cognition such distributed
information requires a mechanism of multiregional function interaction.
(Besthorn et al. ,1994; Locatelli et al., 1998; Babiloni et al. ,2004; Jelles et al.,2008; Sankari and Adeli 2011)
investigated EEG coherence in 16, 17 and 19 electrodes according to 10-20 international system. They estimated
coherence for individual electrode and all neighboring pair electrodes, fronto-parietal, interhemispheric and
intrahemispheric networks, distal and anterior-posterior regions. A decrease in average coherence was seen in
frontal and central regions of theta, alpha and beta bands in AD patients. Moreover, particular coherence was
identified in AD patients with controls. [5] evaluated the functionality of cortical connectivity and synchronization
of regional cortical connectivity in AD patients using EEG coherence measure. The cortical connectivity in AD
declined due to decrease in coherence and have potential to distinguish the AD and control subjects.
Relative power and coherence used by [9] to analyze the amnestic mild cognitive impairment in diabetes. They
investigated alpha coherence and correlations in intra region (parietal) and inter region (right temporo-posterior,
fronto-posterior), interhemispheric (LH, RH), Left posterior (LP)-right posterior (RP) regions in aMIC and control
groups and found significant results in alpha frequency band while other bands exhibits no significant differences.
From the results, it is evident that topographic maps show differences in arousal and activation during EC and EO.
Thus these are not considered as baseline conditions. Moreover, topographic maps using wavelet coherent methods
high distinguish the conditions using phase synchronization properties computed from wavelet transform. Thus
wavelet coherence methods are more robust to analyze the nonlinear dynamical problems. Thus pairwise Wavelet
Coherence in all frequency bands was computed and summarized to see the overall affect EEG in EC and EO during
resting state. The table and topographic maps also depicts the increase and decrease of coherence in inter-
hemispheric and intera-hemispheric regions. Thus the current researches will insight the future researches as a
biomarker for complex and nonlinear physiological markers in EEG with different pathologies.
302 Lal Hussain and Wajid Aziz / Procedia Computer Science 95 (2016) 297 – 302

References

1. Barry, R. J., Clarke, A. R., Johnstone, S. J., Magee, C. A., & Rushby, J. A. (2007). EEG differences between eyes-closed and eyes-open
resting conditions. Clinical Neurophysiology, 118(12), 2765-2773.
2. O’gorman, R. L., Poil, S. S., Brandeis, D., Klaver, P., Bollmann, S., Ghisleni, C., ... & Michels, L. (2013). Coupling between resting
cerebral perfusion and EEG. Brain topography, 26(3), 442-457.
3. Clemson, P. T., Suprunenko, Y. F., Stankovski, T., & Stefanovska, A. (2014). Inverse approach to chronotaxic systems for single-variable
time series. Physical Review E, 89(3), 032904.
4. Sheppard, L. W., Stefanovska, A., & McClintock, P. V. E. (2012). Testing for time-localized coherence in bivariate data. Physical Review
E, 85(4), 046205.
5. Sankari, Z., Adeli, H., & Adeli, A. (2011). Intrahemispheric, interhemispheric, and distal EEG coherence in Alzheimer’s disease. Clinical
Neurophysiology, 122(5), 897-906.
6. Laufs, H., Kleinschmidt, A., Beyerle, A., Eger, E., Salek-Haddadi, A., Preibisch, C., & Krakow, K. (2003). EEG-correlated fMRI of human
alpha activity. Neuroimage, 19(4), 1463-1476.
7. Schiavone, G., Linkenkaer-Hansen, K., Maurits, N. M., Plakas, A., Maassen, B. A., Mansvelder, H. D., ... & van Zuijen, T. L. (2014).
Preliteracy signatures of poor-reading abilities in resting-state EEG. Frontiers in human neuroscience, 8.
8. Murias, M., Swanson, J. M., & Srinivasan, R. (2007). Functional connectivity of frontal cortex in healthy and ADHD children reflected in
EEG coherence. Cerebral Cortex, 17(8), 1788-1799.
9. Bian, Z., Li, Q., Wang, L., Lu, C., Yin, S., & Li, X. (2014). Relative power and coherence of EEG series are related to amnestic mild
cognitive impairment in diabetes. Frontiers in aging neuroscience, 6.